2.20.1.4 Linux Postinstallation Notes

mysql.server can be found in the support-files directory under the MySQL installation directory or in a MySQL source tree. You can install it as /etc/init.d/mysql for automatic MySQL startup and shutdown. See Section 2.18.1.2, “Starting and Stopping MySQL Automatically”.

If MySQL cannot open enough files or connections, it may be that you have not configured Linux to handle enough files.

In Linux 2.2 and onward, you can check the number of allocated file handles as follows:

shell> cat /proc/sys/fs/file-max
shell> cat /proc/sys/fs/dquot-max
shell> cat /proc/sys/fs/super-max

If you have more than 16MB of memory, you should add something like the following to your init scripts (for example, /etc/init.d/boot.local on SuSE Linux):

echo 65536 > /proc/sys/fs/file-max
echo 8192 > /proc/sys/fs/dquot-max
echo 1024 > /proc/sys/fs/super-max

You can also run the echo commands from the command line as root, but these settings are lost the next time your computer restarts.

Alternatively, you can set these parameters on startup by using the sysctl tool, which is used by many Linux distributions (including SuSE Linux 8.0 and later). Put the following values into a file named /etc/sysctl.conf:

# Increase some values for MySQL
fs.file-max = 65536
fs.dquot-max = 8192
fs.super-max = 1024

You should also add the following to /etc/my.cnf:

[mysqld_safe]
open-files-limit=8192

This should enable a server limit of 8,192 for the combined number of connections and open files.

The STACK_SIZE constant in LinuxThreads controls the spacing of thread stacks in the address space. It needs to be large enough so that there is plenty of room for each individual thread stack, but small enough to keep the stack of some threads from running into the global mysqld data. Unfortunately, as we have experimentally discovered, the Linux implementation of mmap() successfully unmaps a mapped region if you ask it to map out an address currently in use, zeroing out the data on the entire page instead of returning an error. So, the safety of mysqld or any other threaded application depends on the gentlemanly behavior of the code that creates threads. The user must take measures to make sure that the number of running threads at any given time is sufficiently low for thread stacks to stay away from the global heap. With mysqld, you should enforce this behavior by setting a reasonable value for the max_connections variable.

If you build MySQL yourself, you can patch LinuxThreads for better stack use. See Section 2.20.1.3, “Linux Source Distribution Notes”. If you do not want to patch LinuxThreads, you should set max_connections to a value no higher than 500. It should be even less if you have a large key buffer, large heap tables, or some other things that make mysqld allocate a lot of memory, or if you are running a 2.2 kernel with a 2GB patch. If you are using our binary or RPM version, you can safely set max_connections at 1500, assuming no large key buffer or heap tables with lots of data. The more you reduce STACK_SIZE in LinuxThreads the more threads you can safely create. Values between 128KB and 256KB are recommended.

If you use a lot of concurrent connections, you may suffer from a feature in the 2.2 kernel that attempts to prevent fork bomb attacks by penalizing a process for forking or cloning a child. This causes MySQL not to scale well as you increase the number of concurrent clients. On single-CPU systems, we have seen this manifest as very slow thread creation; it may take a long time to connect to MySQL (as long as one minute), and it may take just as long to shut it down. On multiple-CPU systems, we have observed a gradual drop in query speed as the number of clients increases. In the process of trying to find a solution, we have received a kernel patch from one of our users who claimed it helped for his site. This patch is available at http://dev.mysql.com/Downloads/Patches/linux-fork.patch. We have done rather extensive testing of this patch on both development and production systems. It has significantly improved MySQL performance without causing any problems and is recommended for users who still run high-load servers on 2.2 kernels.

This issue has been fixed in the 2.4 kernel, so if you are not satisfied with the current performance of your system, rather than patching your 2.2 kernel, it might be easier to upgrade to 2.4. On SMP systems, upgrading also gives you a nice SMP boost in addition to fixing the fairness bug.

We have tested MySQL on the 2.4 kernel on a two-CPU machine and found MySQL scales much better. There was virtually no slowdown on query throughput all the way up to 1,000 clients, and the MySQL scaling factor (computed as the ratio of maximum throughput to the throughput for one client) was 180%. We have observed similar results on a four-CPU system: Virtually no slowdown as the number of clients was increased up to 1,000, and a 300% scaling factor. Based on these results, for a high-load SMP server using a 2.2 kernel, it is definitely recommended to upgrade to the 2.4 kernel at this point.

We have discovered that it is essential to run the mysqld process with the highest possible priority on the 2.4 kernel to achieve maximum performance. This can be done by adding a renice -20 $$ command to mysqld_safe. In our testing on a four-CPU machine, increasing the priority resulted in a 60% throughput increase with 400 clients.

If you see a dead mysqld server process with ps, this usually means that you have found a bug in MySQL or you have a corrupted table. See Section B.5.4.2, “What to Do If MySQL Keeps Crashing”.

To get a core dump on Linux if mysqld dies with a SIGSEGV signal, you can start mysqld with the --core-file option. Note that you also probably need to raise the core file size by adding ulimit -c 1000000 to mysqld_safe or starting mysqld_safe with --core-file-size=1000000. See Section 4.3.2, “mysqld_safe — MySQL Server Startup Script”.